See discussions, stats, and author profiles for this publication at: https://www.researchgate.

net/publication/3481287

Design and Optimization of Printed Spiral Coils for Efficient Transcutaneous

Inductive Power Transmission

Article in IEEE Transactions on Biomedical Circuits and Systems · October 2007

DOI: 10.1109/TBCAS.2007.913130 · Source: IEEE Xplore

CITATIONS READS
616 4,501

2 authors:

Uei-Ming Jow Maysam Ghovanloo

Georgia Institute of Technology Silicon Creations Inc.
30 PUBLICATIONS 2,099 CITATIONS 374 PUBLICATIONS 11,634 CITATIONS

SEE PROFILE SEE PROFILE

All content following this page was uploaded by Maysam Ghovanloo on 29 November 2013.

The user has requested enhancement of the downloaded file.

IEEE TRANSACTIONS ON BIOMEDICAL CIRCUITS AND SYSTEMS, VOL. 1, NO. 3, SEPTEMBER 2007 193

Design and Optimization of Printed Spiral

Coils for Efficient Transcutaneous
Inductive Power Transmission
Uei-Ming Jow, Student Member, IEEE, and Maysam Ghovanloo, Member, IEEE

Abstract—The next generation of implantable high-power neu- inside the eyeball through a 5-mm incision [2], and invasive
roprosthetic devices such as visual prostheses and brain computer brain–machine interfaces are being developed for the 1 3 mm
interfaces are going to be powered by transcutaneous inductive epidural spacing between the outer surface of the brain and the
power links formed between a pair of printed spiral coils (PSC)
that are batch-fabricated using micromachining technology. Op- skull [3].
timizing the power efficiency of the wireless link is imperative to Unlike pacemakers, high power requirements and extreme
minimize the size of the external energy source, heating dissipa- size constraints in aforementioned applications, which belong to
tion in the tissue, and interference with other devices. Previous a broader category of implantable devices known as neuropros-
design methodologies for coils made of 1-D filaments are not com-
prehensive and accurate enough to consider all geometrical as- theses, have prevented the use of batteries as the primary source
pects of PSCs with planar 3-D conductors as well as design con- of energy within the implanted devices. Instead, the battery is
straints imposed by implantable device application and fabrication kept out of the patient’s body and power is delivered wirelessly
technology. We have outlined the theoretical foundation of optimal across the skin through a pair of inductively coupled coils that
power transmission efficiency in an inductive link, and combined
constitute a transformer [4]. In the case of deep brain stimula-
it with semi-empirical models to predict parasitic components in
PSCs. We have used this foundation to devise an iterative PSC tors (DBS), however, the electrodes are kept in place within the
design methodology that starts with a set of realistic design con- neural tissue and the rest of the implant has been displaced to
straints and ends with the optimal PSC pair geometries. We have the chest area where there is more space available [5]. Hence,
executed this procedure on two design examples at 1 and 5 MHz DBS implants can afford to include a primary battery at the cost
achieving power transmission efficiencies of 41.2% and 85.8%, re-
spectively, at 10-mm spacing. All results are verified with simula- of a subcutaneous cable running from the chest across the neck
tions using a commercial field solver (HFSS) as well as measure- to the electrodes on top of the head [6].
ments using PSCs fabricated on printed circuit boards. Even outside of the body, the battery needs to be small, light
Index Terms—Coupling coefficient, implantable microelectronic weight, and have a long lifetime due to portability, aesthetic,
devices, inductive wireless links, neuroprostheses, power transmis- and economic reasons. The external part of the cochlear im-
sion efficiency, telemetry. plants, for example, should fit comfortably behind the user’s
ears. Therefore, the efficiency of the power transmission from
the external battery to the implanted electronic load should be
I. INTRODUCTION
maximized to extend the battery lifetime. Another reason is
IRELESS operation of implantable microelectronic de-
W vices is necessary for clinical applications in order to
reduce the risk of infection and patient discomfort, which can
the limited allowable tissue exposure to the ac magnetic fields,
which can result in excessive heat dissipation if violated [7], [8].
Interference between the RF link and other nearby communica-
result from transcutaneous wires breaching the skin. A wireless tion devices and appliances is also a concern, which is regulated
implant is also safer, more robust, and less likely to be dam- by the Federal Communications Commission (FCC) [9].
aged as a result of broken interconnects. Another important re- Fig. 1 shows a simplified diagram of the transcutaneous
quirement, which has made implantable devices applicable to power flow and various losses along the way as well as the
the treatment of a growing number of ailments and conditions inductive link circuit model schematic. is the primary coil
such as deafness, blindness, and paralysis, is being minimally that is attached to the skin from outside of the body and is
invasive, which is a direct consequence of small size. Cochlear the secondary coil that is implanted under the skin flap often
implants need to be placed inside the 3 6 mm thick temporal with the rest of the implant electronics. In the case of cochlear
bone near the ears [1], retinal implants are expected to be placed implants a pair of permanent magnets, one in the center of
each coil, aligns and holds them together. Coil windings have
Manuscript received April 16, 2007; revised September 25, 2007. This work distributed parasitic resistance and capacitance associated with
was supported in part by the Department of Electrical and Computer Engi- them, which are shown as lumped elements, and
neering at North Carolina State University (NCSU). This paper was recom-
mended by Editor-in-Chief T. S. Lande.
, in Fig. 1(b). Capacitors and are usually
The authors are with the GT Bionics Laboratory, Department of Electrical added to the circuit to form a pair of resonance LC-tank circuits
and Computer Engineering, Georgia Institute of Technology, Atlanta, GA 30308 with and , respectively.
USA (e-mail: [email protected]; [email protected]). The largest power loss usually takes place in the primary coil
Color versions of one or more of the figures in this paper are available online
at http://ieeexplore.ieee.org. parasitic resistance followed by and the power condi-
Digital Object Identifier 10.1109/TBCAS.2007.913130 tioning blocks within on the implant side. The latter may
1932-4545/$25.00 © 2008 IEEE
194 IEEE TRANSACTIONS ON BIOMEDICAL CIRCUITS AND SYSTEMS, VOL. 1, NO. 3, SEPTEMBER 2007

within the epidural space. They can also conform to the outer
body or brain surface curvature if fabricated on thin flexible
substrates such as polyimide [23]. Rigid hermetically sealed
PSCs can also be fabricated on silicon chips or low temperature
co-fired ceramics (LTCC) using micromachining techniques
[24], [25].
When dealing with PSCs, the design procedures that are
devised for coils made of 1-D filament wires are only good
for finding the optimal lumped values for the inductive link
circuit models, shown in Fig. 1(b). They are not compre-
hensive enough to predict the intertwined effects of the dis-
tributed inductive, resistive, and capacitive elements on power
efficiency when dealing with 3-D PSCs. Further, the previous
methods do not take into account many of the design constraints
that are dictated by implantable device application or PSC fab-
Fig. 1. (a) Power flow diagram in a transcutaneous inductive power trans-
mission link from the external battery to implanted electronics. (b) Simplified rication technology. Therefore, a new design methodology is
schematic diagram of the inductive link with lumped equivalent circuit ele- needed to find all the parameters associated with the geometries
ments. of the planar PSCs and maximize . Since the closed form
equations governing the electromagnetic fields and parasitics
of 3-D conductors rapidly become complicated, it becomes
however deem to be more important because it is within the necessary to take advantage of the available simulation tools
tissue [7]. There is also power loss within the external ac source, as part of the design methodology and validate the design out-
, which is often an efficient class-E power amplifier [10]. The come.
power loss within the surrounding tissue can be ignored if the Even though PSCs have been used quite widely in radio fre-
operating frequency is chosen below 20 MHz or so quency identification (RFID) applications, there has been little
[11]. Therefore, it is fair to say that the overall power trans- research on optimizing the geometry of these coils in order to
mission efficiency, , is often dominated by maximize their power transmission efficiency [12], [19]. This
, which we will focus on throughout the rest of is probably because most RFID applications are extremely low
this paper. It should also be pointed out that all efficiencies are power (data storage and retrieval) and less size constrained com-
functions of the inductive link operating frequency, , pared to high power neuroprosthetic applications. Shah et al.
which is also known as the power carrier [12]. derived some general guidelines from experimental data [23].
Design and optimization of efficient inductive power trans- However, they stopped short of a well-defined design method-
mission links have been well studied over the last few decades ology supported by theoretical foundation and validated by sim-
[4], [13]–[16]. More recently there have been a number of pub- ulations and experiments.
lications proposing new approaches to the same optimization In this paper, we provide a procedure for designing the ge-
problems [17]–[21]. A summary of these studies in Section II ometries of a pair of lithographically defined single layer planer
shows that the coils mutual inductance and quality factor, spiral coils that optimizes their mutual inductance and quality
and , are the most important param- factors in a way that is maximized. The design procedure
eters affecting . It is also well known from electromagnetics also takes into account the major design constraints imposed
that , , and are dependant on the coils geometries, which by regulations, application, and PSC fabrication technology. We
in general terms can also include their relative distance, orien- limited our design to single layer PSCs even though many fabri-
tation, and number of turns. Hence, we can conclude that the cation processes allow using multilayer conductors. Because the
coils geometries are the most important factors in defining the PSC internal parasitic capacitance of multilayer PSCs is much
link power efficiency. larger than single layer and it has a significant effect on reducing
Previous designs have been mostly based on coils made of the self-resonance frequency (SRF) of the PSC, which in turn
filament wires in the form of a single or multiple individually limits the power carrier frequency, as can be seen in our design
insulated strands. The latter type, which is also known as Litz examples.
wire, helps in reducing the skin-loss resistive effects espe- We have used closed-form equations in MATLAB (Math-
cially at higher frequencies by increasing the circumferential Works, Natik, MA) in combination with HFSS (Ansoft, Pitts-
area of the wire [22]. Wire-wound coils, however, cannot be burgh, PA) to find the optimal coil geometries. However, any
batch-fabricated or reduced in size without the use of sophis- other field solver can also be used for this purpose [26]. We have
ticated machinery. Therefore, the shape and geometry of the also fabricated a number of PSCs, designed through this proce-
coils used in the next generation of high power implantable dure, on FR4 PCB and characterized them to validate our PSC
devices is more likely to be planar and lithographically defined design technique.
in one or multiple layers on rigid or flexible substrates similar In the next section we have briefly reviewed the theoretical
to printed circuit boards (PCB). Printed spiral coils (PSC) offer design equations followed by the design procedure and simula-
more flexibility in optimizing their geometry and aspect ratio, tion results in Section III. Measurement results are included in
deeming them attractive for implantation under the skin or Section IV followed by concluding remarks.
JOW AND GHOVANLOO: DESIGN AND OPTIMIZATION OF PRINTED SPIRAL COILS 195

where is the skin depth, is the permeability of space, and

is the relative permeability of the conductor [31]. Thus, the
quality factor of the PSC without its parasitic capacitance can
be found by substituting (1)–(5) in .
We adopted a semi-empirical approach in finding the para-
sitic capacitance of the PSC using a combination of theoretical,
simulation, and measurement results [32]. A parallel plate par-
asitic capacitor forms between the spiral conductor sidewalls,
shown in Fig. 2(b) cross section, along the spiral gap, shown in
Fig. 2(a). The length of the gap is slightly shorter than the length
of conductor and can be found from

(6)

There are two types of insulating materials affecting this ca-

pacitance. One is air or the coating insulator that fills the gap
between adjacent traces. The other is the PSC substrate, which
could be ceramic, polyimide, or FR4. Therefore, we divided
Fig. 2. Geometrical parameters of a square-shaped printed spiral coil. (a) Top into and components
view. (b) Cross section of a pair of parallel conductors showing the parasitic
capacitor components within the insulator and substrate.
(7)

where and are the relative dielectric constants of the

II. INDUCTIVE COUPLING AND POWER EFFICIENCY THEORY coating and substrate materials, respectively. We empirically
found that in the case of air and FR4, . From
A. Planar Spiral Coils Quality Factor insulator characteristic tables, [33].
The ratio between the magnetic flux generated by a current Considering that is in series with and is in par-
carrying conductor and the current passing through it indicates allel with both [see Fig. 1(b)], the overall impedance and quality
its self-inductance. Several closed-form equations have been factor of a PSC can be found from (8) and (9), respectively
proposed to approximate in PSCs [27]–[30]. We adopted (1)
from [30] for square coils shown in Fig. 2(a) (8)

(1)

(2) for small or low (9)

where is the number of turns and are the outer and inner
diameters of the coil, respectively, , and is B. Coils Mutual Inductance
a parameter known as fill factor, which changes from 0, when all A PSC can be considered a set of concentric single-turn coils
the turns are concentrated on the perimeter like filament coils, with shrinking diameters, connected in series. Therefore, once
to 1, when the turns spiral all the way to the center of the coil. we find the mutual inductance between a pair of single-turn coils
To find the total parasitic dc resistance of the PSC, we need in parallel planes, the overall can be found by summing the
to know the length of the conductive trace , resistivity of the partial mutual inductance values between every turn on one coil
conductive material , and its thickness and all the turns on the other coil. Using Maxwell equations,
between a pair of parallel circular single-turn coils at radii
(3) and can be found from
(4)

where and are the line width and spacing, respectively.

The skin effect will increase the coil ac resistance at higher
frequencies and should be taken into account (10)

where is the relative distance between the two coils, is

the permeability of the medium, and is the lateral misalign-
(5) ment [15], [34]. and are the Bessel functions of the zeroth
196 IEEE TRANSACTIONS ON BIOMEDICAL CIRCUITS AND SYSTEMS, VOL. 1, NO. 3, SEPTEMBER 2007

and first order, respectively. For perfectly aligned coaxial coils, On the secondary side, there is a power division between ,
where , (10) can be simplified to which is wasted as heat, and . To find , can be trans-
formed to its parallel equivalent at resonance [37]
(11)
(18)
(12)
Since and both receive the same voltage across the
tank, it is easy to verify that
where and are the complete elliptic integrals of the
first and second kind, respectively [35]. By adding the partial (19)
mutual inductances between every two turns on a PSC pair
Therefore, from (17) and (19), we can demonstrate the factors
(13) that are most effective in the link power efficiency [21]

(20)
where is a factor dependent on the shape of the PSC. Even
though the area of a square-shaped coil with a side length of In PSCs, most aforementioned parameters are interrelated.
is 27% larger than a circular coil with equal diameter, we empiri- For example, increasing the number of turns for each coil
cally found that between a pair of square-shaped PSCs is only without changing the coil outer diameter can increase and .
10% higher than a pair of similar circular PSCs. Thus, 0.95, However, it may decrease by increasing due to increased
1.0, and 1.1 for a pair of hexagonal, circular, and square-shaped and reduced . Therefore, there are optimal PSC geometries
PSCs with equal diameters, respectively. that would maximize . Parameters such as substrate loss and
eddy current also play a role, which are not included in detail
C. Inductive Link Efficiency due to their small effects. Another parameter that has not been
It can be shown mathematically that the highest voltage gain included is the output resistance, which is small in class-E
and efficiency across an inductive link can be achieved when amplifiers and should be considered in [10].
both LC-tanks are tuned at the link operating frequency,
. This is the first set of constraints III. DESIGN PROCEDURE
in choosing and values in Fig. 1(b). The power delivered to
An iterative design procedure has been depicted in this
the primary LC-tank, , divides between , which is wasted
section which starts with a set of design constraints and initial
as heat, and the secondary loading, which is reflected onto the
values, and ends with the optimal PSC pair geometries. HFSS
primary through . To simplify the efficiency equations, is
simulations have been used for fine tuning and verifying the
usually normalized w.r.t. and by defining as the coils
values suggested by theoretical calculations. MATLAB-pro-
coupling coefficient [36]
duced 3-D curves demonstrate the changes in when
geometrical parameters change in a wide range. By sweeping
(14)
numerous parameters included in (1)–(20), even those that are
In practice, the secondary coil is always loaded as shown in not directly utilized in our optimization procedure, such as PSC
Fig. 1 by . The loaded secondary quality factor at resonance misalignments in (10), the designer can get a good sense of
can be found from [12] how to make necessary compromises and changes to achieve a
satisfactory performance.
Step 1: Applying Design Constraints: There are a set of pa-
(15) rameters affecting the wireless link efficiency which are im-
posed by other factors related to the implantable device appli-
cation or fabrication technology. The former usually defines the
overall size constraints depending on where the implant will
where if . If the nominal is
be located inside the body. The latter indicates the minimum
known, it is also possible to find the value that maximizes
size features that result in acceptable yield in manufacturing.
by differentiating (15) w.r.t.
Table I summarizes a number of these parameters and their ex-
emplary values in our design for a cortical visual prosthesis [23],
[38], [39].
(16) The power carrier frequency has a significant effect on the
efficiency and optimal geometry of the PSCs. In order to demon-
strate its effects, in the next few steps we have followed the it-
Considering the above points, Ko et al. showed that [4]
erative procedure for two different operating frequencies, 1 and
5 MHz. The difference between the final design outcomes shows
the importance of and the efficacy of the proposed design
(17) method.
JOW AND GHOVANLOO: DESIGN AND OPTIMIZATION OF PRINTED SPIRAL COILS 197

TABLE I
DESIGN CONSTRAINS IMPOSED BY APPLICATION AND
FABRICATION TECHNOLOGY

Depending on whether a chip or magnet is going to be placed in the center

of the PSC or not.
1-oz copper on FR4 printed circuit board.

Step 2: Initial Values: A set of initial values need to be se-

lected before starting the iterative optimization process. In our
earlier work [40], we found that the optimal fill factor for a pair
of identical spiral filament coils is . For mm,
this leads to mm. The optimal value for depends
on the coils relative distance, , which is usually not fixed and
has a certain range. Therefore, depending on the application and
design constraints, can be chosen to be the nominal distance
where the coils are for most of the time during normal operation
or the maximum distance, which indicates the worst case power
transmission.
In a single-turn circular filament coil with radius , it can be
shown that the field strength, , along the axis at distance
from the coil center can be found from [12]

(21)

Differentiating (21) w.r.t. , shows that for , will

be maximized. Therefore, a good choice for would be

(22)

In our design example, mm and mm.

The number of turns in a PSC is defined based on , , and
, as opposed to 1-D filament coils, which have no limits. A
larger number of turns generally results in larger , , , and
. We initiate and with the minimum values permitted
by the fabrication technology, , consid- Fig. 3. Optimizing the size and fill factor of the primary PSC while w and s are
ering the fact that this choice may not be optimal. Increasing at their minimum. (a)  versus d and ' at 1 MHz. (b)  versus d and
' at 5 MHz. (c) Calculated, simulated, and measured efficiency versus d .
in the subsequent steps is likely to improve since it can
reduce . However, since increasing results in decreasing
or , it should only be considered if the SRF of the PSC, For MHz, it can be seen in Fig. 3(a) that the best choice
, approaches . for would be 48 mm and almost flattens for .
Step 3: Size and Fill Factor of Primary PSC: To optimize This indicates that the turns very close to the center of the PSC
component of in the primary PSC, and need to be consid- do not help with improving the efficiency. If we set ,
ered together according to (17). Plugging the initial values from then mm and . Assuming mm
Steps 1 and 2 in (1)–(20), we found while sweeping and results in and , which should improve
in a wide range around their initial values. Since and are once and secondary PSC are optimized. Similarly, when
constants in this step, changing directly affects according to MHz, the 3-D surface in Fig. 3(b) reaches its maximum at
mm and flattens for . Setting
(23)
results in mm, , and .
198 IEEE TRANSACTIONS ON BIOMEDICAL CIRCUITS AND SYSTEMS, VOL. 1, NO. 3, SEPTEMBER 2007

Fig. 3(c) compares the simulated and calculated efficiencies

versus . Calculated results at low frequency (1 MHz) are very
accurate because conductive traces are narrow (150 m) and
close to filament wires. Narrow traces also reduce the effects of
eddy currents. At higher frequencies (5 MHz), the effects of PSC
parasitic capacitance become more significant and (5) overesti-
mates for larger diameters. The peak efficiency is signifi-
cantly higher at 5 MHz due to higher PSC quality factors. How-
ever, rapidly drops as approaches its SRF. The hollow
circles at mm are the measured efficiency values,
using PSC11 and PSC21 in Table III to form a link, which vali-
date calculation and simulation results.
Step 4: Fill Factor and Line Width of Secondary PSC: After
temporarily resolving the geometry of the primary PSC in
Step 3, we focus on the secondary PSC. Considering that
is already specified based on the implant size in Step 1, we
sweep around its nominal value and increase from
, while calculating from (1)–(20). It can be seen from
Fig. 4(a) that in our exemplary design at 1 MHz the peak is
slightly below . Since the fabrication technology does
not allow any thinner conductor, we chose and
, which corresponds to turns. This yields
. Substituting in (2) gives mm, which
is smaller than . If we choose mm to leave room
for a chip or magnet in the center of the secondary PSC, ,
, and reduce to 0.43, 20, and 34.8%, respectively. At 5
MHz, the secondary PSC geometries that result from Step 4
are mm, m, mm, and ,
which yield , as shown in Fig. 4(b).
Fig. 4(c) compares cross-sections of Fig. 4(a) and (b) curves
at with simulated values of versus using
HFSS. It can be seen that the theoretical and simulated values
are very close with maximum efficiencies of 33.2% and 75.12%
at 1 and 5 MHz, respectively.
Step 5: Size and Line Width of Primary Coil: In this step
we return back to the primary PSC and increase the conductor
width, , towards its optimal value while providing room for
this change by increasing . Increasing is likely to increase
by reducing and increasing . However, it also requires
larger and may reduce by deviating from (22) or by reducing
. Fig. 5(a) shows how an optimal point with is
reached at 1 MHz when m and mm. Sub-
stituting these values in (23) along with mm from Step 3
gives . Following the same steps at 5 MHz results in
the 3-D curve shown in Fig. 5(b) with a peak at m
and mm, yielding and . Fig. 4. Optimizing the fill factor and conductor line width of the secondary
Fig. 5(c) compares cross-sections of Fig. 5(a) and (b) at PSC. (a)  versus w and ' at 1 MHz. (b)  versus w and ' at 5 MHz. (c)
mm, with simulated values of versus . Once again Comparison between calculated and simulated  versus w when ' = 0:43.
there is a good agreement between calculated and simulated
values especially in the midrange. The difference is higher at
the higher end of these curves for wider conductors because the Step 6: Iteration by Going to Step 3: The PSC geometries
effects of eddy currents become more prominent, particularly from Step 5 significantly improve compared to the initial
at higher frequencies. HFSS considers those effects while our values. However, further improvement is possible by iterating
equations in Section II do not. The simulation versus calcula- through Steps 3–5. Iterations can continue until improvement
tion difference also increases at the lower end of these curves in per iteration is less than 0.1%. Table II depicts the final
for narrower conductors because of the inaccuracies of (7) and optimized values of the design examples in Table I after several
the effects of fringe and substrate parasitic capacitors, which are iterations. Fig. 6 summarizes the iterative PSC design procedure
ignored in calculations but not in simulations. in a flowchart.
JOW AND GHOVANLOO: DESIGN AND OPTIMIZATION OF PRINTED SPIRAL COILS 199

TABLE II
OPTIMIZED GEOMETRIES FOR PSC PAIR DESIGN EXAMPLES

Fig. 5. Optimizing the size and conductor line width of the primary PSC. (a)
 versus d and w at 1 MHz. (b)  versus d and w at 5 MHz. (c) Com-
parison between calculated and simulated  versus w when d = 70 mm.
Fig. 6. Iterative PSC design flowchart.

IV. MEASUREMENT RESULTS TABLE III

Table III shows the geometries and specifications of five SPECIFICATIONS OF THE PSCS USED IN MEASUREMENTS
square-shaped PSCs that were fabricated using 1-oz copper on
FR4 substrates. PSC11 14 were used as primaries and formed
an inductive link with PSC21 as the secondary. PSC pairs were
mounted on Plexiglas sheets using plastic screws, aligned, and
held in parallel. Even though these geometries were not exactly
the same as the optimal values found in Section III, they could
validate the accuracy of our calculations and HFSS simulations,
which can be extended to the proposed optimization procedure.
For efficiency measurements we used two setups. First, a net-
work analyzer (Agilent E5071B), shown in Fig. 7(a) connected
200 IEEE TRANSACTIONS ON BIOMEDICAL CIRCUITS AND SYSTEMS, VOL. 1, NO. 3, SEPTEMBER 2007

Fig. 8. Comparison between calculated (Section II), simulated (HFSS), and

Q f
measured (network analyzer) variations of versus for PSC13 in Table III.

Fig. 7. (a) Experimental setup for measuring the S-parameters between a pair
of PSCs using a network analyzer. (b) 3-D PSC models constructed in the HFSS
full-wave electromagnetic field simulator.

to PSC1x and PSC21 to measure the S-parameters. The S-pa-

rameters are converted to Z-parameters [41], which can then be
substituted in (14)–(20) to find . Second, a circuit similar to
Fig. 1(b) was used with a function generator (HP 8111A) driving
PSC1x at resonance frequency, . An oscilloscope (Tektronix
DPO 4034) was also connected to one of the PSCs at a time
measuring its current and voltage waveforms without violating
Fig. 9. Efficiency variations with coils relative distance for PSC21 in Table III
the wireless link isolation. The transmitted and received powers as secondary and (a) PSC11 and PSC12 as primary at 1 MHz and (b) PSC13
can then be easily calculated by multiplying these waveforms. and PSC14 as primary at 5 MHz.
Fig. 7(b) shows 3-D PSC models constructed with the same
dimensions as the fabricated ones in the HFSS full-wave elec-
tromagnetic field simulator. Fig. 8 shows variations versus Fig. 10(a) and (b) compares versus for fabricated coils
frequency for PSC13. This is also a comparison between cal- in Table III at mm. Once again there is a good agree-
culated (Section II), simulated (HFSS), and measured (network ment between calculated, simulated, and measured values par-
analyzer) values. It can be seen that the results are close and ticularly at low frequencies. Efficiency is maximized around
all suggest that the highest can be achieved around 5 MHz. 4 5 MHz mainly because of the high of the PSCs at these
At low frequencies increases as . At high frequencies, frequencies, as shown in Fig. 8.
however, the skin and loading effects degrade according to
(5) and (15), respectively. V. CONCLUSION
Fig. 9(a) and (b) compares versus at 1 and 5 MHz, We have devised an iterative design procedure to optimize
respectively. It can be seen that calculated, simulated, and mea- the geometries of a pair of printed spiral coils used in im-
sured results are in agreement. PSC12 and PSC14, which is plantable microelectronic devices to maximize the inductive
closer to the optimal value in Table II, provide better efficiencies power transmission efficiency between them. Unlike previous
particularly at the nominal mm. design procedures which are mostly based on 1-D filament
JOW AND GHOVANLOO: DESIGN AND OPTIMIZATION OF PRINTED SPIRAL COILS 201

the proposed iterative PSC design procedure can maximize the

power transfer efficiency and provide insight on its changes with
PSC geometries. We are now using this PSC design technique in
development of a multicarrier inductive wireless link [36], [42].

ACKNOWLEDGMENT
The authors would like to thank Prof. G. Lazzi and his group
for their help with the measurements.

REFERENCES
[1] G. M. Clark, Cochlear Implants: Fundamentals and Applications.
New York: Springer-Verlag, 2003.
[2] M. S. Humayun et al., “Visual perception in a blind subject with
a chronic microelectronic retinal prosthesis,” Vis. Res., vol. 43, pp.
2573–2581, Nov. 2003.
[3] R. R. Harrison et al., “A low-power integrated circuit for a wireless
100-electrode neural recording system,” IEEE J. Solid-State Circuits,
vol. 42, no. 1, pp. 123–133, Jan. 2007.
[4] W. H. Ko, S. P. Liang, and C. D. F. Fung, “Design of radio-frequency
powered coils for implant instruments,” Med. Biol. Eng. Comput., vol.
15, pp. 634–640, 1977.
[5] H. A. Shill and A. G. Shetter, “Reliability in deep brain stimulation,”
IEEE Trans. Device Materials Reliab., vol. 5, no. 3, pp. 445–448, Sep.
2005.
[6] M. Ghovanloo, “Switched-capacitor based implantable low-power
wireless microstimulating systems,” in IEEE Intl. Symp. Circuits Syst.,
May 2006, pp. 2197–2200.
[7] G. Lazzi, “Thermal effects bioimplants,” IEEE Eng. Med. Biol. Mag.,
pp. 75–81, Sep. 2005, vol..
[8] IEEE Standard for Safety Levels With Respect to Human Exposure to
Radio Frequency Electromagnetic Fields, 3 kHz to 300 GHz, IEEE
Standard C95.1, 1999.
Fig. 10. Efficiency variations with power carrier frequency for PSC21 in [9] Federal Communication Commission, Wireless Medical Telemetry
Table III as secondary located 10 mm away from (a) PSC11 and PSC12 as [Online]. Available: http://www.wireless.fcc.gov/services/index.htm?
primary and (b) PSC13 and PSC14 as primary. job=service_home&id=wireless_medical_telemetry
[10] N. O. Sokal and A. D. Sokal, “Class-E-A new class of high-efficiency
tuned single-ended switching power amplifiers,” IEEE J. Solid-State
Circuits, vol. SSC-10, no. 2, pp. 168–176, Jun. 1975.
wires, the new method takes into account the effects of the [11] J. C. Lin, “Computer methods for field intensity predictions,” in CRC
distributed inductive, resistive, and capacitive components Handbook of Biological Effects of Electromagnetic Fields, C. Polk
and E. Postow, Eds. Boca Raton, FL: CRC Press, 1986, ch. 2, pp.
simultaneously on the coils quality factors and coupling coef- 273–313.
ficient and changes the PSC geometries such that the net result [12] K. Finkenzeller, RFID Handbook: Fundamentals and Applications in
would improve the overall power transfer efficiency. Further, Contactless Smart Cards and Identification, 2nd ed. Hoboken, NJ:
Wiley, 2003.
since this procedure shows how power efficiency changes when [13] N. N. Donaldson and T. A. Perkins, “Analysis of resonant coupled coils
various PSC geometrical parameters vary in a wide range, it in the design of radio frequency transcutaneous links,” Med. Biol. Eng.
allows the designer to make proper compromises to accom- Comput., vol. 21, no. 5, pp. 612–627, Sep. 1983.
[14] C. M. Zierhofer and E. S. Hochmair, “High-efficiency coupling-in-
modate for other non-geometrical design considerations. The sensitive transcutaneous power and data transmission via an inductive
procedure starts with applying the design constraints that are link,” IEEE Trans. Biomed. Eng., vol. 37, no. 7, pp. 716–722, Jul. 1990.
usually imposed by the intended application of the implantable [15] C. M. Zierhofer and E. S. Hochmair, “Geometric approach for
coupling enhancement of magnetically coupled coils,” IEEE Trans.
device and PSC fabrication technology. Initial values are also Biomed. Eng., vol. 43, no. 7, pp. 708–714, Jul. 1996.
chosen for all PSC geometrical parameters that affect the power [16] W. J. Heetderks, “RF powering of millimeter and submillimeter-sized
neural prosthetic implants,” IEEE Trans. Biomed. Eng., vol. 35, no. 5,
efficiency. The next step is optimizing size and fill factor of the pp. 323–327, May 1988.
primary PSC, followed by fill factor and conductor width in the [17] C. R. Neagu, H. V. Jansen, A. Smith, J. G. E. Gardeniers, and M. C.
secondary PSC. Then we return back to the primary side and Elwanspoek, “Characterization of a planar microcoil for implantable
microsystems,” Sens. Actuat. A, vol. 62, pp. 599–611, 1997.
optimize its size and conductor width. This iteration repeats [18] G. A. Kendir et al., “An optimal design methodology for inductive
until the power efficiency reaches its maximum level for the power link with class-E amplifier,” IEEE Trans. Circuits Syst. I, Reg.Pa-
selected operating frequency and the PSCs relative distance. pers, vol. 52, no. 5, pp. 857–866, May 2005.
[19] S. C. Q. Chen and V. Thomas, “Optimization of inductive RFID tech-
The PSC design procedure was validated by applying it to de- nology,” in Proc. IEEE Int. Symp. Electron. Environ., May 2001, pp.
sign examples for a cortical visual prosthesis at 1 and 5 MHz. 82–87.
We constructed finite element analysis models in HFSS using [20] R. R. Harrison, “Designing efficient inductive power links for im-
plantable devices,” in IEEE Int. Symp. Circuits Syst., May 2007, pp.
the optimal geometries and simulated them. We also fabricated 2080–2083.
several PSC pairs with different geometries on FR4 and mea- [21] M. W. Baker and R. Sarpeshkar, “Feedback analysis and design of RF
sured their coupling coefficient and power efficiency. All calcu- power links for low-power bionic systems,” IEEE Trans. Biomed. Cir-
cuits Syst., vol. 1, no. 1, pp. 28–38, Mar. 2007.
lation, simulation, and measurement results were in close agree- [22] F. E. Terman, Radio Engineers Handbook. New York: McGraw-Hill,
ment in the desired range of parameters and demonstrated that 1943.
202 IEEE TRANSACTIONS ON BIOMEDICAL CIRCUITS AND SYSTEMS, VOL. 1, NO. 3, SEPTEMBER 2007

[23] M. R. Shah, R. P. Phillips, and R. A. Normann, “A study of printed [41] D. M. Pozar, Microwave Engineering, 2nd ed. New York: Wiley,
spiral coils for neuroprosthetic transcranial telemetry applications,” 1998, ch. 4.
IEEE Trans. Biomed. Eng., vol. 45, no. 7, pp. 867–876, Jul. 1998. [42] M. Ghovanloo and S. Atluri, “A wideband power-efficient inductive
[24] J. A. Von Arx and K. Najafi, “A wireless single-chip telemetry-pow- wireless link for implantable microelectronic devices using multiple
ered neural stimulation system,” in Proc. IEEE Int. Solid-State Circuits carriers,” IEEE Trans. Circuits Syst. I, Reg. Papers, vol. 54, no. 10, pp.
Conf. Dig., Feb. 1999, pp. 214–215. 2211–2221, Oct. 2007.
[25] S. Kim, M. Wilke, M. Klein, M. Toepper, and F. Solzbacher, “Electro-
magnetic compatibility of two novel packaging concepts of an induc-
tively powered neural interface,” in Proc. 3rd Int. IEEE EMBS Conf.
Neural Eng., May 2007, pp. 434–437.
[26] FastHenry2 [Online]. Available: http://www.fastfieldsolvers.com/ Uei-Ming Jow (S’07) received the B.E. degree in electrical engineering
[27] F. W. Grover, Inductance Calculations: Working Formulas and Ta- from Tatung University, Taiwan, R.O.C., and the M.S. degree in electronic
bles. New York: Van Nostrand, 1946. engineering from National Taiwan University of Science and Technology,
[28] H. Greenhouse, “Design of planar rectangular microelectronic induc- Taiwan, R.O.C., in 1999 and 2001, respectively. He was working towards the
tors,” IEEE Trans. Parts, Hybrids, Packag., vol. PHP-10, no. 2, pp. Ph.D. degree at the North Carolina State University, Raleigh, since fall 2006
101–109, Jun. 1974. and has recently moved to the Georgia Institute of Technology, Atlanta.
[29] A. Balakrishnan, W. D. Palmer, W. T. Joines, and T. G. Wilson, “The From 2001 to 2006, he joined Industrial Technology Research Institute
inductance of planar structures,” in Proc. 8th Annu. Appl. Power Elec- (ITRI), Hsinchu, Taiwan, R.O.C., and worked in Electronics Research and
tron. Conf. Expo., Mar. 7–11, 1993, pp. 912–921. Service Organization as a RF engineer. He was involved in analysis and
[30] S. S. Mohan, M. del Mar Hershenson, S. P. Boyd, and T. H. Lee, design of electromagnetic compatibility for high-speed digital circuit as well
“Simple accurate expressions for planar spiral inductances,” IEEE J. as embedded RF circuits packaging technology. His main research interests
Solid-State Circuits, vol. 34, no. 10, pp. 1419–1424, Oct. 1999. are neural and bionic implants, integrated analog circuit design, and wireless
[31] H. A. Wheeler, “Formulas for the skin effect,” Proc. IRE, vol. 30, no. implantable biomedical systems.
9, pp. 412–424, Sep. 1942.
[32] R. Garg and I. J. Bahl, “Characteristics of coupled microstriplines,”
IEEE Trans. Microw. Theory Tech., vol. 27, no. 7, pp. 700–705, Jul.
1979.
[33] L. W. Ritchey, “A survey and tutorial of dielectric materials used in the Maysam Ghovanloo (S’00–M’04) was born in 1973. He received the B.S. de-
manufacture of printed circuit boards,” CircuiTree Mag., Nov. 1999. gree in electrical engineering from the University of Tehran, Tehran, Iran, in
[Online]. Available: http://www.circuitree.com/ 1994 and the M.S. (Hons.) degree in biomedical engineering from the Amirkabir
[34] C. Akyel, S. Babic, and S. Kincic, “New and fast procedures for calcu- University of Technology, Tehran, Iran, in 1997. He also received the M.S. and
lating the mutual inductance of coaxial circular coils (circular coil-disk Ph.D. degrees in electrical engineering from the University of Michigan, Ann
coil),” IEEE Trans. Magn., vol. 38, no. 5, pp. 2367–2369, Sep. 2002. Arbor, in 2003 and 2004, respectively. His Ph.D. research was on developing a
[35] M. Soma, D. G. Galbraith, and R. L. White, “Radio-frequency coils wireless microsystem for Micromachined neural stimulating microprobes.
in implantable devices: Misalignment analysis and design procedure,” From 1994 to 1998, he worked part-time at IDEA Inc., Tehran, Iran, where he
IEEE Trans. Biomed. Eng., vol. BME-34, pp. 276–282, Apr. 1987. participated in the developing a modular patient care monitoring system. In De-
[36] M. Ghovanloo and G. Lazzi, “Transcutaneous magnetic coupling of cember 1998, he founded Sabz-Negar Rayaneh Co. Ltd., Tehran, Iran, to man-
power and data,” in Wiley Encyclopedia of Biomedical Engineering, ufacture physiology and pharmacology research laboratory instruments. In the
M. Akay, Ed. Hoboken, NJ: Wiley, Apr. 2006. summer of 2002, he was with the Advanced Bionics Inc., Sylmar, CA, working
[37] T. H. Lee, The Design of CMOS Radio-Frequency Integrated Circuits, on the design of spinal-cord stimulators. From 2004 to 2007, he was an As-
2nd ed. Cambridge, U.K.: Cambridge University Press, 2004. sistant Professor at the Department of Electrical and Computer Engineering,
[38] M. Ghovanloo and K. Najafi, “A modular 32-site wireless neural stim- North Carolina State University, Raleigh, where he founded and directed the
ulation microsystem,” IEEE J. Solid-State Circuits, vol. 39, no. 12, pp. NC Bionics Laboratory. In June 2007, he joined the faculty of Georgia Institute
2457–2466, Dec. 2004. of Technology, Atlanta, where he is currently an Assistant Professor in the De-
[39] M. Ghovanloo and K. Najafi, “A wireless implantable multichannel partment of Electrical and Computer Engineering.
microstimulating system-on-a-chip with modular architecture,” IEEE Dr. Ghovanloo has received awards in the operational category of the 40th
Trans. Neural Syst. Rehab. Eng., vol. 15, no. 3, pp. 449–457, Sep. 2007. and 41st DAC/ISSCC student design contest in 2003 and 2004, respectively.
[40] S. Atluri and M. Ghovanloo, “A wideband power-efficient induc- He has served as a Technical Reviewer for major IEEE and IoP journals in the
tive wireless link for implantable biomedical devices using multiple areas of circuits, systems, and biomedical engineering. He is a member of Tau
carriers,” in Proc. IEEE Int. Symp. Circuits Syst., May 2006, pp. Beta Pi, Sigma Xi, and IEEE Solid-State Circuits, IEEE Circuits and Systems,
1131–1134. and IEEE Engineering in Medicine and Biology Societies.

View publication stats